JPH0585056B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a text editing device for creating and editing Japanese texts. In particular, it is used in industrial fields where Japanese word processors have traditionally been used.

(Prior Art) Conventionally, Japanese word processors have been popular as devices for creating and editing Japanese sentences. The Japanese word processor mainly stores and displays character strings containing Kanji, Kana, and Kanji of Japanese sentences inputted through Kana-Kanji conversion. It then accesses the stored strings containing kanji and kana to perform editing functions such as inserting, deleting, moving, copying, and replacing strings.

(Problem to be solved by the invention) Conventional Japanese word processors store Japanese sentences only as character strings, so they cannot be inserted or
Editing functions such as deleting, moving, copying, and replacing cannot be performed on a character-by-character basis.

On the other hand, when editing Japanese text, inserting and
The unit to be processed by deletion, movement, copying, replacement, etc. is usually a structural unit of a Japanese sentence, such as a sentence, clause, or word, or a semantically coherent unit. On the other hand, with conventional Japanese word processors, the user reconsiders the unit of target sentences, clauses, words, etc. as a range of character strings, and then calculates the first character position of the range and the last character position of the range. Editing operations such as deleting, moving, and copying are performed by specifying . At this time, it is cumbersome to reconsider the target sentence, clause, word, etc. as a range of character strings, and this also causes the cursor to move back and forth between the character positions you are trying to specify, or the specified position Sometimes I shift the ``by'' by one character. Therefore, it is possible to specify and edit the target in units of sentences, clauses, words, etc.
It is desirable to be able to move the cursor in units of phrases, words, etc.

On the other hand, an English text editing device that allows you to specify units such as words, sentences, and paragraphs for editing was published by Smith et al. in Byte Magazine (Smith, D., lrby, Co. , Verplank.B.
and Harslem, E.: “Designing the Star User
Interface”, Byte, Vol. 7, No. 4, 1982). In English, words are separated by spaces, so a string of characters between spaces is considered a “word” and a period is used. A string of characters between (.) can be treated as a "sentence". However, in the case of Japanese sentences, it is not customary to separate words with spaces, so the English method cannot be used as is.

Therefore, in the case of Japanese sentences, there is a method to consider character strings of the same character type as words (Ito, Tanaka: "Editor using Bitmap display and mouse",
``Information Processing'' Vol. 25, No. 8, 1984) and a method of considering the input unit as a clause when inputting kana-kanji conversion (Japanese Unexamined Patent Publication No. 1982-69829 ``Document Creation Device''), but these This method is only an approximation and lacks accuracy. For example, when a method is adopted in which character strings of the same glyph type are regarded as words, in the sentence ``Since then he pursued a new method.'' Words such as ``method'' and ``pursuit'' can be specified successfully, but `` "Since then" and "He" cannot be separated; "Since then" is "He".
, and "new" can only specify "new". For example, when using a method in which the input unit for kana-kanji conversion input is regarded as a phrase,
When entering ``letter'', you may enter ``hand'', ``paper'', and ``wo'' separately, or you may enter ``letter ni'' and later correct ``ni'' to ``wo''. In this case, ``letter'' cannot be treated as a single phrase.

Furthermore, in conventional character string editing, it is not possible to specify the part of speech of a word, resulting in incorrect processing. For example, when trying to replace the noun ``repetition'' with the noun ``repetition,'' the conventional method not only replaces the noun ``repetition'' that appears as a noun in the sentence (for example, ``repetition'' in example 1 below) , even the ``repetition'' (for example, ``repetition'' in Example 2 below) of the verb's conjunctive form is subject to substitution.
This has the disadvantage that the result is a sentence that does not make sense (for example, Example 2 below becomes Example 3).

(Example 1) ``Repetition'' of the same word has the effect of emphasis.

(Example 2) “Repeating” the same words over and over again.

(Example 3) “Repeating” the same words over and over again.

Furthermore, conventionally, kana-kanji conversion has been mainly used to input character strings containing kanji and kana in Japanese sentences. In this kana-kanji conversion, the pronunciation of a Japanese sentence is analyzed, the units of sentences, clauses, and words, and the part of speech of the word are determined, and then the text is converted into a string containing kanji and kana. However, since this kana-kanji conversion is used as an input method, it is usually
Information obtained through analysis, such as units of sentences, clauses, and words, and parts of speech of words, is discarded once the conversion process is completed, and is not used effectively. In JP-A No. 59-69829 ``Document Creation Device'', input unit information is used as a clause unit, but it is also possible to use the overall sentence structure such as word units, parts of speech of words, and sentence units. There is nothing planned.

The purpose of the present invention is to eliminate the drawbacks of conventional character string editing as described above, and to provide accurate and easy-to-operate editing based on sentence structure information such as units of sentences, clauses, and words, and parts of speech of words. It is an object of the present invention to provide a text editing device that makes it possible and effectively utilizes the analysis results of kana-kanji conversion.

(Means for Solving the Problems) The present invention provides a text editing device for editing Japanese texts, which includes an input means for inputting pronunciations of Japanese texts, and a dictionary storage means for storing a word dictionary. and kana-kanji conversion, which refers to the word dictionary and determines sentence structure information such as sentence/clause/word units and part of speech of each word for the reading kana of the Japanese sentence, and converts it into a character string containing kanji and kana. means, a text storage means for storing the kanji-kana-mixed character string and the edited kanji-kana-mixed character string of the Japanese text output from the kana-kanji conversion means; A sentence analysis means that determines sentence structure information such as the units of sentences, clauses, and words and parts of speech of each word for the stored character string containing kanji and kana, and the sentence analysis means or the kana-kanji conversion means. a structure specifying means for specifying the text structure information; and a structure specifying means for specifying the text structure information; a structure position extraction means for extracting a position; and a reanalysis determination means for monitoring sentence structure information written from the kana-kanji conversion means to the sentence structure storage means and determining whether or not to perform reanalysis by the sentence analysis means. This is a text editing device characterized by comprising:

(Example) The structure and operation of the present invention will be explained using the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of the text editing device of the present invention.

In FIG. 1, input means 1 inputs a sequence of pronunciations of Japanese sentences, inputs editing specifications such as insertion, deletion, movement, copying, and replacement, and inputs instructions to move the cursor position. Usually a keyboard is used. When an editing command, a movement command, etc. is input, the input means 1 sends it to the text editing control means 9, and when a pronunciation string is input, it notifies the text editing control means 9 that the pronunciation kana has been input. , the reading kana string is sent to the kana-kanji conversion means 10.
Pronunciations can be input using kana keys or Roman characters using alphabet keys.

The dictionary storage means 4 stores a word dictionary. I C
Memory, magnetic disk devices, etc. are used. A word dictionary registers at least the notation, pronunciation, and part of speech for Japanese words.
It is referenced by the kana-kanji conversion means 10 and the text analysis means 5. When referred to by the kana-kanji conversion means 10, the reading kana is the key, and when referred to from the text analysis means 5, the notation is the key.

The kana-kanji conversion means 10 refers to the word dictionary stored in the dictionary storage means 4 for the pronunciation kana string of the Japanese sentence inputted from the input means 1, and converts it into a sentence/kanji.
It determines sentence structure information such as the unit of clauses and words and the part of speech of each word, and converts it into a character string containing kanji and kana. The kana-kanji conversion means 10 is a publicly known means, and is realized, for example, as in "Separation of solid text and kana-kanji conversion" (Makino et al., Journal of the Information Processing Society of Japan, Vol. 20, No. 4, 1979). can. However, a function will be added to output information such as units of sentences, clauses, words, parts of speech of words, etc., which were conventionally used only internally in Kana-Kanji conversion (addition of such a function will be done by those skilled in the art). can be easily achieved). The kana-kanji conversion means 10 sends the determined character string containing kanji and kana to the sentence storage means 2, and sends sentence structure information such as sentence/clause/word units and parts of speech of each word to the sentence structure storage means 6. .

The text storage means 2 stores the kanji-kana-mixed character string sent from the kana-kanji conversion means 10 and the kanji-kana-mixed character string edited by the text editing control means 9. IC memory, magnetic disk devices, etc. are used.

The display means 3 displays the kanji/kana mixed character string stored in the text storage means 2, a cursor, and the like. A CRT display or the like is used.

The sentence analysis means 5 refers to the word dictionary stored in the dictionary storage means 4 and selects a range of characters specified by the sentence editing control means 9 from among the character strings containing kanji and kana stored in the sentence storage means 2. For character strings containing kanji and kana, sentence structure information such as units of sentences, clauses, and words, as well as the part of speech of each word, is determined. The sentence analysis means 5 is a known means, for example, "Japanese dictionary memory and automatic segmentation of Japanese sentences" (Nagao et al.
It can be realized as shown in "Information Processing" Vol. 19 No. 6, 1978).

The sentence structure storage means 6 is a kana-kanji conversion means 10.
and the sentence structure information of the character string containing kanji and kana characters determined by the sentence analysis means 5 are stored. IC memory, magnetic disk devices, etc. are used. As the sentence structure information, specifically, the position of sentence boundaries, the position of clause boundaries, the position of word boundaries, the part of speech of each word, etc. are stored.

The structure specifying means 7 specifies sentence structure information.
A keyboard or the like is used, and the input means 1 can also be used. Specifically, the contents to be specified are as follows.
These include units such as sentences, clauses, and words, their number, and types of parts of speech. For example, to specify units such as sentences, clauses, and words, keys such as "sentence,""clause," and "word" may be prepared. The number can be specified using the numeric keypad, or by increasing or decreasing the number using the "+" or "-" keys. The part of speech of a word can be specified by determining a code corresponding to the part of speech and inputting that code.

The structure position extraction means 8 refers to the sentence structure storage means 6 and extracts a position in the sentence based on the sentence structure information. Specifically, upon receiving the current cursor position and the designation by the structure designation means 7, the sentence structure storage means 6 is searched and the corresponding character position or range is extracted. Or, given a string range,
Extract the part-of-speech code corresponding to that range.

The reanalysis determination means 11 is a kana-kanji conversion means 10.
If the sentence structure information in the sentence structure storage means 6 output from the sentence structure storage means 6 changes due to editing, or
When a new sentence is input and new sentence structure information is written, it is determined whether reanalysis by the sentence analysis means 5 is to be performed. For example, if the sentence structure information output from the kana-kanji conversion means 10 to the sentence structure storage means 6 satisfies any of the following conditions, it is determined that the sentence analysis means 5 should reanalyze it.

(A) It is an unregistered word.

(B) One-letter nouns are consecutive.

(A) not only means a word that is not in the word dictionary, but also a single character input (not as a word, but as one of a group of kanji characters), or a code input. Characters (entered by specifying a character code), character strings entered without conversion (entered by specifying "no conversion" without going through the process of converting kana character strings to kanji), etc. including. When it is determined that reanalysis is to be performed, the reanalysis determination means 11 further determines the range to be reanalyzed. The range to be reanalyzed may include, for example, the part that satisfies the conditions (A) and (B), and the clauses before and after the part. The determination result of the reanalysis determination means 11 and the range for reanalysis are sent to the text editing control means 9.

The text editing control means 9 exchanges information with each of the above-mentioned means and realizes editing functions such as input, insertion, deletion, movement, copying, and replacement. The conventional text editing device is composed of only an input means 1, a kana-kanji conversion means 10, a text storage means 2, a display means 3, and a text editing control means 9. The text editing control means 9 performs, for example, the following processing. conduct.

(1) When a kana character string is input from the input means 1, the kana-kanji conversion 10 is started, and the character string containing kanji and kana as a result of the kana-kanji conversion is inserted and written into the text storage means 2. The writing position is the character position corresponding to the cursor position.

(2) The character string containing kanji and kana characters stored in the text storage means 2 is displayed on the display means 3. The cursor is also displayed at the same time.

(3) The position of the cursor displayed on the display means 3 is changed in accordance with the cursor movement command sent from the input means 1.

(4) Receive the first character position and last character position of the range sent from input means 1, and display the character string between them in a different display method from the normal character string (for example, by displaying it in reverse video) ).

(5) When a "delete" command is sent from the input means 1, the character string corresponding to (4) is deleted from the text storage means 2 (accordingly, the display on the display means 3 is also changed). ).

(6) When a "move" command is sent from the input means 1, the character string corresponding to (4) is deleted from the text storage means 2 and stored in the buffer in the text editing control means 9 ( Along with this, the display on the display means 3 is also changed).

(7) When a "copy" command is sent from the input means 1, the character string corresponding to (4) is copied from the text storage means 2 to the buffer in the text editing control means 9.

(8) When the movement destination/copy destination position is sent from the input means 1, the content of the buffer in the text editing control means 9 is changed to correspond to the movement destination/copy destination position.
It is inserted and written at the position of the text storage means 2 (accompanying this, the display on the display means 3 is also changed).

(9) When a "replace" command, a character string to be replaced, and a character string to be replaced are sent from the input means 1, the character string to be replaced is searched from the text storage means 2, and the character string to be replaced is found. If so, replace it with the replacement string.

In addition to the processes (1) to (9) above, the text editing control means 9 in the text editing apparatus of this embodiment includes a text analysis means 5, a text structure storage means 6, a structure specification means 7, a structure position extraction means 8 , controls the operation of the reanalysis determination means 11 to perform the following processing.

(10) When a kana character string is input from the input means 1 and (1) is performed, the kana-kanji conversion means 10
A character string containing kanji and kana as a result of converting a kana character string is written into a text storage means 2, and text editing structure information corresponding to the character string containing kanji and kana is written into a text structure storage means 6.

(11) When a character string in the text storage means 2 is changed due to deletion, movement, copying, etc., the position of the character string is notified to the text analysis means 5, and the text analysis means 5 is instructed to analyze the character string. and the contents of the sentence structure storage means 6 are changed (the contents of the sentence structure storage means 6 always correspond to the contents of the sentence storage means 2).

(12) Specify units such as “sentences”, “clauses”, and “words” from structure specifying means 7 (including specifying the number of units)
is sent, the current cursor position and its unit specification are sent to the structure position extraction means 8, and the first and last character positions of the corresponding character string range are received from the structure position extraction means 8. . Next, process (4) is performed.

(13) At the time of (9), if the part-of-speech code is sent from the structure specifying means 7, the position of the replaced character string found from the sentence storage means 2 is sent to the structure position extraction means 8, and the structure position The corresponding part-of-speech code is received from the extraction means 8. Structure designation means 7
When the part-of-speech code input from the part-of-speech code and the part-of-speech code received from the structural position extraction means 8 match, the same replacement process as in (9) is performed.

(14) When the reanalysis determination means 11 sends the determination result that reanalysis is to be performed and the range to be reanalyzed, the text analysis means 5 is notified of the reanalysis range, and the text analysis means 5 is notified of the reanalysis range. The character string is analyzed and the contents of the sentence structure storage means 6 are changed.

Next, the operation of this embodiment will be explained using an example.

First, it is assumed that the following kana character string is input by the input means 1.

In the middle of the day, the wordpress is blowing as if it were a house that performs a variety of exercises. Things are progressing on the waterfront. The Word Producer listens to the writings and uses them.

The input means 1 notifies the text editing control means 9 that the kana character string has been input, and sends the kana character string to the kana-kanji conversion means 10. The kana-kanji conversion method is
Analyze the kana string and convert it to the following string containing kanji and kana.

Conventionally, word processors have been widely used as devices for creating and editing texts. Word processors are becoming cheaper. A word processor stores and displays text.

The kana-kanji conversion means 10 writes this character string containing kanji and kana into the text storage means 2, and further writes the text structure information resulting from the analysis into the text structure storage means 6.

FIG. 2a shows the contents of the sentence storage means 2 for this example, and FIG. 2b shows the contents of the sentence structure storage means 6 for this example. The positions of the elements in the matrices a and b in FIG. 2 correspond. In Figure 2b, the position where s is written indicates that that position is the first character position of a sentence (and also the first character position of a clause), and the position where p is written indicates that position. indicates that it is the first character position of the clause. Positions where other values are written indicate the first character positions of the word, and the values represent the part of speech of the word as follows.

Name...Noun verb...Verb complement...Auxiliary verb head...Prefix suffix...Suffix auxiliary...Particle point...Punctuation mark However, the method of realizing the sentence structure storage means 6 is not limited to that shown in FIG. 2b. For example, instead of creating a matrix corresponding to a as shown in Figure 2b, sentence initial positions (0,0), (3,7), (5,8), phrase initial positions (0,0), (0,3 ), (1,3), (1,6),...
,
(7,5), word (0,0, name), (0,2, points),
(0,3, name), ..., (7,5, motion), (7,9, points)
This can also be achieved by enumerating the positions.

Here, it is assumed that an operation is performed to delete the sentence ``The price of word processors is decreasing.''

The conventional method of deleting a string is to first position the cursor at the "wa" position, press the "start" key,
Next, position the cursor at the "." position and click "End".
Specify the statement to be deleted by pressing the key (then press the "delete" key to designate the command to be deleted).

In contrast, in the method of this embodiment, by pressing the "Sentence" key while the cursor is at an arbitrary position in the sentence "Word processors are becoming cheaper." , the sentence to be deleted can be specified (then pressing the "Delete" key is the same).

How each component operates at that time will be described below.

Suppose that the cursor is at the ``valence'' position and the ``sentence'' key is pressed. The "sentence" key is input from the structure specifying means 7 and sent to the sentence structure control means 9. The text editing control means 9 controls the current cursor position (4,
7) and the designated structural unit "sentence" are sent to the structural position extraction means 8.

The structure position extraction means 8 refers to the sentence structure storage means 6, and extracts the closest s before the position (4, 7) (representing the beginning position of the sentence as described above) and the position after the position (4, 7). Find the nearest s. As a result, positions (3, 7) and (5, 8) are obtained. Then, the position (3, 7) is sent as the start position of the character string range, and the position (5, 7) one character ahead as the end position is sent to the text editing control means 9. In response to this, the text editing control means 9 edits the character string from position (3, 7) to position (5, 7) as follows.
The display means 3 is instructed to change the display method from that of other character strings, for example, to display the character string in reverse video.

Subsequently, the "delete" key is input from the input means 1 and sent to the text editing control means 9. In response to this, the text editing control means 9 deletes the character string from position (3, 7) to position (5, 7) in the text storage means 2, and deletes the character string from position (5, 8) onwards. Move up to position (3,7). Then, the text analysis means 5 is further activated, and the text structure information corresponding to the content of the text storage means 2 is transferred to the text structure storage means 6.
write to.

However, when deleting a sentence unit as in this example, it is not necessarily necessary to redo the sentence analysis.
8) and subsequent steps can be realized by simply moving up to position (3, 7). By specifying a range as a conventional string,
If one end of the range falls in the middle of a phrase or word, it is necessary to redo the sentence analysis.

As a result, the contents of the sentence storage means 2 and the sentence structure storage means 6 become as shown in FIGS. 3a and 3b.

In this way, the deletion of "sentence" units has been realized, but when it is "sentence" units, s in the sentence structure storage means 6 is deleted.
Instead of searching for the position of p, it is sufficient to search for the position of p, and when the unit is a "word", it is sufficient to search for the position containing the part of speech. Structure position extraction means 8
However, after returning the character string range corresponding to the structural unit to the text editing control means 9, the processing of the text editing control means 9 is similar to conventional processing, so it not only "deletes" the structural unit but also deletes the structural unit. "Move" and "Copy" units
It is clear that "replacement" and "cursor movement" can be realized in the same way.

Furthermore, when inputting a sentence using kana-kanji conversion, it is not necessarily input in units of sentences or phrases. For example, it would be fine if all the words ``I wrote a letter'' were input, but after inputting ``letter,'' I was confused as to what to add to the particle, and then only the single character ``wo'' was input. Sometimes. Alternatively, input may be made by cutting the information into small pieces, such as "hand" or "paper."

In such a case, since the reanalysis determination means 11 monitors the sentence structure information written from the kana-kanji conversion means 10 to the sentence structure storage means 6, it is necessary for the reanalysis determination means 11 to perform reanalysis. It is determined that this is the case, and the text analysis means 5 re-analyzes the text. In the above example, if "letter" and "wo" are input, "wo" will be an unregistered word, and if "te" and "paper" are input, "te" and "paper" will be one character. Since it is a series of nouns, it is determined that reanalysis is necessary. If there are sentences before and after ``letter'', the range of reanalysis will include the phrases before and after ``letter''. When reanalysis is performed by the sentence analysis means 5, "letter" is recognized as "letter" (noun) and "wo" (particle), and correct sentence structure information is formed.

(Effects of the Invention) As explained above, according to the present invention, editing is performed in units of sentence structure such as "sentences,""clauses," and "words," or editing is performed based on part-of-speech information. becomes possible.

In the case of conventional character string editing, it was necessary for humans to reconsider what they were thinking in their heads as a range of character strings, and then perform the editing operation. The object can be specified naturally, and the operability of the text editing device is greatly improved.

The conventional method of specifying character strings of the same character type,
Compared to the method of specifying the input unit for Kana-Kanji conversion input, the units of words, phrases, and sentences that can be specified using this method are much more accurate, making editing easier for users. It becomes extremely easy to specify the target.

Furthermore, not only the structural unit of a sentence but also the part of speech of a word can be used as information during editing, which improves the accuracy of editing such as substitution compared to the past.

In addition, the analysis results of kana-kanji conversion can be used effectively without being thrown away, and since it is equipped with a re-analysis judgment means, it is not necessary to input phrases into small pieces during kana-kanji conversion. However, it is possible to reformulate the correct sentence structure.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, and FIGS. 2a and 2b and 3a and 3b are diagrams showing the contents of the text storage means 2 and the text structure storage means 6 in the embodiment. It is. In the figure, 1... input means, 2... text storage means, 3... display means, 4... dictionary storage means, 5... text analysis means, 6... sentence structure storage means, 7... structure specification means, 8... structure position extraction means , 9... Text editing control means, 10... Kana-kanji conversion means, 11... Reanalysis determination means.

Claims (1)

[Claims] 1. A text editing device that edits a Japanese text includes an input means that can input the pronunciation of the Japanese text, a dictionary storage means that stores a word dictionary, and a text editing device that edits the Japanese text by referring to the word dictionary. A kana-kanji conversion means that determines sentence structure information such as the units of sentences, clauses, and words and parts of speech of each word for the reading kana and converts it into a character string containing kanji and kana, and the Japanese language output from the kana-kanji conversion means. text storage means for storing a character string containing kanji and kana of a word sentence and an edited character string containing kanji and kana; a sentence analysis means for determining sentence structure information such as the units of clauses and words and parts of speech of each word; and a sentence structure storage means for storing the sentence structure information determined by the sentence analysis means or the kana-kanji conversion means. , a structure specifying means for specifying sentence structure information, a structure position extraction means for referring to the sentence structure storage means and extracting a position in a sentence based on the sentence structure information specified by the structure specifying means, and the kana-kanji character. 1. A text editing device comprising: reanalysis determination means for monitoring text structure information written from conversion means to said text structure storage means and determining whether or not to perform reanalysis by said text analysis means.